Hi ,
I posted a few considerations about fialover implementation , to the list ,
but i don't recived any feed back , well lets see what we have :
But i write a bunch of perl scripts and now I use them in production and
very happy with the result .
Here what I have :
Baiscly what I want is an sql cluster (composed of min 2 servers )
that can survive a crash and still working . (we all want this )
so i set up a very simple replication between a master and a slave
and I use an dynamic dns to locate slave && master .
like this :
master.sql.corp.com 192.168.1.100
slave1.sql.corp.com 192.168.1.101
slaveN.sql.corp.com 192.168.1.10N
I have a wrapper that send all selects to the slaves and updates to the
master
i have an advanced watchdog that check if all sqls are up , and if :
SLAVE FAIL :
I simply updater the DNS to slave-1 and all applications still working
the dns update is very quick (1sec) .
MASTER FAIL :
I simply update the DNS to the NEXT slave for the master and next slave
for
the slave (to avoid all selects on the master)
I run change master to on all slaves , and the party keep going
this works from 2 to N servers ,
I can setup an opensource project , and put all watchdog/dns update /setup
stuff
in it if any body is intrestead in this .
I run this for 2 month and all goes very very well i'm very happy with this
.
My cluster is :
5 pc , and I use it for a web cluster (10 pc) so I use 1 slave for 2 httpds
.
For the moment after an server fail , i sync it back using a simple script ,
the watch dog don't do it because i dont want a buggy server coming in and
out
the party .
I would be very happy to release all implementation .
Very best
Constantin Bogomolnyi
----- Original Message -----
From: "Bryan Coon" <[EMAIL PROTECTED]>
To: <[EMAIL PROTECTED]>
Sent: Wednesday, March 14, 2001 6:10 PM
Subject: Failover and Replication
> Hi,
>
> We are implementing a linux cluster, and rather than get gouged by oracle
> (in performance and licensing) would muuuuuch rather use MySQL. I
searched
> through the archives and found several references to failover and
> replication but still have some questions.
>
> If I understand correctly, replication creates a working copy of mysql on
> another server, but it looks like the algorithm to automatically change
> slave to master if the master dies will not be added until mysql 4.x (from
> 11.4 in the manual).
>
> Is this still the case? Im not sure when that was written, perhaps there
is
> news on this front? I also did not find a expected release date
>
> Does anyone have any stories/suggestions/experiences in regards to rolling
> our own monitoring script?
>
> Thanks!
> Bryan
>
> ---------------------------------------------------------------------
> Before posting, please check:
> http://www.mysql.com/manual.php (the manual)
> http://lists.mysql.com/ (the list archive)
>
> To request this thread, e-mail <[EMAIL PROTECTED]>
> To unsubscribe, e-mail
<[EMAIL PROTECTED]>
> Trouble unsubscribing? Try: http://lists.mysql.com/php/unsubscribe.php
>
>
---------------------------------------------------------------------
Before posting, please check:
http://www.mysql.com/manual.php (the manual)
http://lists.mysql.com/ (the list archive)
To request this thread, e-mail <[EMAIL PROTECTED]>
To unsubscribe, e-mail <[EMAIL PROTECTED]>
Trouble unsubscribing? Try: http://lists.mysql.com/php/unsubscribe.php
